Electron beams are increasingly coming into use in the radiation treatment of cancer, because their rapid dose falloff minimizes irradiation of critical healthy tissues beyond the target volume. Unfortunately, for treatment planning purposes it is not yet possible to calculate the absorbed dose distribution precisely, particularly in the presence of tissue inhomogeneities and body curvature. In penetrating the body, high-energy electrons suffer a great many collisions and follow a tortuous, although mainly forward, path. It has only been recently that a number of investigators have begun to apply Fermi-Eyges multiple scattering theory to the problem of electron dosimetry, recognizing that this theory describes the behaviour of the electron beam as a reasonable first approximation, at least in a homogeneous or horizontally layered medium. It is the purpose of the proposed research to develop and apply Fermi-Eyges theory in a way which allows the accurate calculation of electron dose distributions in the presence of tissue inhomogeneities and body curvature using directly the data from computerized tomography (CT) scans. The goal is to achieve an accuracy of 5%, and preferably of 3%, in the practical calculation of electron dose in the most complicated situations likely to be encountered clinically. The method of the proposed research is to develop the Fermi-Eyges theory to a high accuracy, to simplify it mathematically into a practical computational algorithm, and to refine and verify the algorithm through experiments. The key to the theoretical work is the concept of representative electron path, and in addition it will be necessary to include in the computational algorithm a number of secondary effects which are ignored by Fermi-Eyges theory. For the experimental work, four phantoms providing progressively more complicated configurations of inhomogeneities will be use; the dose distribution throughout the phantom will be recorded on radiographic film, and thermoluminescent dosimetry (TLD) will be used to maintain the accuracy of the film dosimetry. CT scans will be used to describe the more complicated phantom configurations to the computational algorithm.